Duke Compsci 220 / ECE 252
Advanced Computer Architecture I
Prof. Alvin R. Lebeck
Unit 0: Introduction
Slides developed by Amir Roth of University of Pennsylvania with sources
that included University of Wisconsin slides by Mark Hill, Guri Sohi, Jim
Smith, and David Wood.
Slides enhanced by Milo Martin, Mark Hill, Alvin Lebeck, Dan Sorin, and
David Wood with sources that included Profs. Asanovic, Falsafi, Hoe,
Lipasti, Shen, Smith, Sohi, Vijaykumar, and Wood
Compsci 220 / ECE 252 (Lebeck): Introduction
1
Instructors
• Prof: Alvin Lebeck
Office: D308 LSRC
Hours: Tues 1:30 – 2:30, Wed 2:00-3:00 or by appointment (email)
email: alvy@cs.duke.edu
• Teaching Assistant (s)
Vali Pistol
Office: D339
Hours: TBD or by appointment
email: pistol@cs.duke.edu
John Ingalls
Hours: TBD or by appointment
Email: john.ingalls@duke.edu
Compsci 220 / ECE 252 (Lebeck): Introduction
2
Course Components
• Reading Materials
 Computer Architecture: A Quantitative Approach by Hennessy and
Patterson, 4th Edition
 Recent research papers (on course website)
• Homework
 5 or 6 homework assignments, performed in groups of 2
• Term Project
 Groups of 2 or 3
• Exams
 Midterm and final exam
Compsci 220 / ECE 252 (Lebeck): Introduction
3
Administrative (Grading)
• 25% Homeworks (must be done in pairs w/ statement)
 4 to 6 Homeworks
 Late < 1 day = 50%
 Late > 1 day = zero
• 50% Examinations (Midterm 20% + Final 30%)
• 25% Research Project (work in groups of 3)
 No late term projects
• Academic Misconduct
 University policy will be followed strictly
 Zero tolerance for cheating and/or plagiarism
• This course requires hard work.
Compsci 220 / ECE 252 (Lebeck): Introduction
4
Administrative (Continued)
• Midterm Exam: In class (75 min) closed book
• Final Exam: (3 hours) closed book
• Big class, if you are going to drop, please do so early.
Compsci 220 / ECE 252 (Lebeck): Introduction
5
Administrative (Continued)
•
Course Web Page




•
http://www.cs.duke.edu/courses/cps220/fall10
Lectures posted there shortly before class (pdf)
Homework posted there
General information about course
Blackboard
 Posting Grades only
• Course Forum
 Google groups – you will receive invite later this week
 Use it to
1. read announcements/comments on class or homework,
2. ask questions (help),
3. communicate with each other.
Compsci 220 / ECE 252 (Lebeck): Introduction
6
Reading
•
H&P
 Chapter 1
• Papers
 Power a First-Class Architectural Design Constraint
 A Case for Energy-Proportional Computing
• After those:
•
Start: Appendix A (pipelining)
Compsci 220 / ECE 252 (Lebeck): Introduction
7
What you should Expect from Course
• Things NOT to expect in this course:
 100% of class = me lecturing to you
 Homework sets and exams where every question is either quantitative or
has a single correct answer
• Things to expect in this course:




Active discussions/arguments about architecture ideas
Essay questions
Being asked to explain, discuss, defend, and argue
Questions with multiple possible answers
Compsci 220 / ECE 252 (Lebeck): Introduction
8
Expected Background
• Who’s in this class?
 PhD, MS, undergrad from Comp Sci, ECE, & other
 From all over the world…
• Courses you should have taken already
 Basic architecture (Compsci 104 / ECE 152 or equivalent)
 Programming (our simulator is in C)
 Basic OS (Compsci 110 / ECE 153) — not critical, but helpful
• Topics you should remember fondly - I will not cover these
in detail in this course
 Instruction sets, computer arithmetic, assembly programming, I/O
Compsci 220 / ECE 252 (Lebeck): Introduction
9
Term Project
• This is a semester-long research project
 Do not expect to do whole thing in last week…
 I will likely define a default project, but many students choose to pursue
their own ideas
 Project proposals due Thursday, October 14
• You may “combine” this project with a project from
another class, but you MUST consult with me first
• You must absolutely, positively reference prior work
 Do not copy text from any source (paper, book, internet, etc.)
 Please ask me if you have ANY questions
 Not knowing != valid excuse
Now on to computer architecture …
Compsci 220 / ECE 252 (Lebeck): Introduction
10
What is Computer Architecture?
• “Computer Architecture is the science and art of selecting
and interconnecting hardware components to create
computers that meet functional, performance and cost
goals.” - WWW Computer Architecture Page
• An analogy to architecture of buildings…
Compsci 220 / ECE 252 (Lebeck): Introduction
11
What is Computer Architecture?
The role of a building architect:
Construction
Materials
Steel
Concrete
Brick
Wood
Glass
Plans
Design
Goals
Function
Cost
Safety
Ease of Construction
Energy Efficiency
Fast Build Time
Aesthetics
Compsci 220 / ECE 252 (Lebeck): Introduction
Buildings
Houses
Offices
Apartments
Stadiums
Museums
12
What is Computer Architecture?
The role of a computer architect:
“Technology”
Logic Gates
SRAM
DRAM
Circuit Techniques
Packaging
Magnetic Storage
Flash Memory
Manufacturing
Plans
Design
Goals
Function
Performance
Reliability
Cost/Manufacturability
Energy Efficiency
Time to Market
Computer
PCs
Servers
PDAs
Mobile Phones
Supercomputers
Game Consoles
Embedded
Three important differences: age (~60 years vs thousands), rate of change,
automated mass production (magnifies design)
Compsci 220 / ECE 252 (Lebeck): Introduction
13
Design Goals
• Functional
 Needs to be correct
 What functions should it support (Turing completeness aside)
• Reliable




Does it continue to perform correctly?
Hard fault vs transient fault
Google story - memory errors and sun spots
Space probe vs PC reliability
• High performance
 “Fast” is only meaningful in the context of a set of important tasks
 Not just “Gigahertz”
 Impossible goal: fastest possible design for all programs
Compsci 220 / ECE 252 (Lebeck): Introduction
14
Design Goals
• Low cost
 Per unit manufacturing cost (wafer cost)
 Cost of making first chip after design (mask cost)
 Design cost (huge design teams, why? Two reasons…)
• Low power




Energy in (battery life, cost of electricity)
Energy out (cooling and related costs)
Static vs dynamic power, sleep modes, peak vs average
Cyclic problem, very much a problem today
• Challenge: balancing the relative importance of these goals
 And the balance is constantly changing
Compsci 220 / ECE 252 (Lebeck): Introduction
15
Constant Change
“Technology”
Logic Gates
SRAM
DRAM
Circuit Techniques
Packaging
Magnetic Storage
Flash Memory
Goals
Function
Performance
Reliability
Cost/Manufacturability
Energy Efficiency
Time to Market
Applications/Domains
PCs
Servers
PDAs
Mobile Phones
Supercomputers
Game Consoles
Embedded
• Absolute improvement, different rates of change, cyclic
• Better computers help design the next generation (CAD)
Compsci 220 / ECE 252 (Lebeck): Introduction
16
Rapid Change
1971–1980 1981–1990 1991–2000
Transistors (M)
2008
2015
0.01–0.1
0.1–1
1–100
300-1000
10000?
Clock (MHz)
0.2–2
2–20
20–1000
3500
10000?
MIPS
<0.2
0.2–20
20–2000
16000
100000?
• Exciting: perhaps the fastest moving field … ever
 Processors vs. cars
o 1985: processors = 1 MIPS, cars = 60 MPH
o 2000: processors = 500 MIPS, cars = 30,000 MPH?
 What if cars crashed as often as computers?
• Computation is driving most, if not all, fields
 Natural and physical sciences, social sciences, journalism, etc.
Compsci 220 / ECE 252 (Lebeck): Introduction
17
First Microprocessor
• Connect a few transistors together to make…
• Intel 4004
• 1971 (first microprocessor)
•
•
•
•
•
•
Compsci 220 / ECE 252 (Lebeck): Introduction
4-bit data
2300 transistors
10 mm PMOS
108 KHz
12 V
13 mm2
18
More Recent Microprocessor
• Or a few more to form…
• Intel Pentium4 + HT
• 2003
•
•
•
•
•
•
Compsci 220 / ECE 252 (Lebeck): Introduction
32/64-bit data
55M transistors
0.90 mm CMOS
3.4 GHz
1.2 V
101 mm2
19
By end of course, this will make sense!
• Pentium 4 specifications: what do each of these mean?
 Technology
o 55M transistors, 0.90 mm CMOS, 101 mm2, 3.4 GHz, 1.2 V
 Performance
o 1705 SPECint, 2200 SPECfp
 ISA
o X86+MMX/SSE/SSE2/SSE3 (X86 translated to RISC uops inside)
 Memory hierarchy
o 64KB 2-way insn trace cache, 16KB D$, 512KB–2MB L2
o MESI-protocol coherence controller, processor consistency
 Pipeline
o 22-stages, dynamic scheduling/load speculation, MIPS renaming
o 1K-entry BTB, 8Kb hybrid direction predictor, 16-entry RAS
o 2-way hyper-threading
Compsci 220 / ECE 252 (Lebeck): Introduction
20
A recently announced processor
• Sun Rock












Compsci 220 / ECE 252 (Lebeck): Introduction
16 Microcores
1-2 threads per core
8 x 32KB L1D$
4 x 32KB L1I$
2MB L2$
16MB L3$
48 GB/s memory B/W
2.1 GHz
296 mm^2
321 million transistors
250W power
Available 2H2009
21
Layers of abstraction
• Architects need to understand computers at many levels





Instruction Set Architecture
Microarchitecture
Systems Architecture
Technology
Applications
• Good architects are “Jacks of most trades”
Compsci 220 / ECE 252 (Lebeck): Introduction
22
Instruction Set Architecture
• Hardware/Software interface
 Software impact
o support OS functions
OP
• restartable instructions
• memory relocation and protection
o
R2
R3
imm
a good compiler target
• simple
• orthogonal
o
R1
OP
M1 R1
M2 R2
imm2
Dense
• Improve memory performance
M3 R3
imm2
 Hardware impact
o admits efficient implementation
• across generations
o
admits parallel implementation
• no ‘serial’ bottlenecks
 Abstraction without interpretation
Compsci 220 / ECE 252 (Lebeck): Introduction
23
Microarchitecture
• Implement instruction set
• Register-transfer-level (RTL) design
• Exploit capabilities of technology
Instr.
Cache
PC
 locality and concurrency
• Iterative process
 generate proposed architecture
 estimate cost
 measure performance
IR
Compsci 220 / ECE 252 (Lebeck): Introduction
C
deep pipelining
multiple issue
dynamic scheduling
branch prediction/speculation
Regs
A




B
• Still emphasis is on overcoming
sequential nature of programs
24
System-Level Design
•
•
•
Design at the level of processors,
memories, and interconnect.
More important to application
performance, cost and power than
CPU (core) design
Feeds and speeds
 constrained by IC pin count,
module pin count, and signaling
rates
•
System balance
 for a particular application
•
P
400MHz
Dual Issue
16Bytes x
200MHz
SW
Display
Net
I/O
Disk
Driven by
 performance/cost goals
 available components (cost/perf)
 technology constraints
M
Compsci 220 / ECE 252 (Lebeck): Introduction
M
M
M
25
Large-system example
•
•
•
•
•
Google Project 02: Oregon/Washington border
Cheap electricity: Columbia river
Cheap cooling: Columbia river
Cheap b/w: surplus optic network from tech-boom era
Google, Yahoo, and Microsoft
Compsci 220 / ECE 252 (Lebeck): Introduction
26
Technology Trends
• Processor (SRAM)
 Density: ~30%, Speed: ~20%
• Memory (DRAM)
 Density: ~60%, Speed: ~4%
• Disk
 Density: ~60%, Speed: ~10%
• Changing quickly and with respect to each other!!
 Fundamentally changes design
 Different tradeoffs
• Exciting: constant re-evaluation and re-design
Compsci 220 / ECE 252 (Lebeck): Introduction
27
Shaping Force: Applications/Domains
• Another shaping force: applications
 Applications and application domains have different requirements
o Domain: group with similar character
 Lead to different designs
• Scientific: weather prediction, genome sequencing
 First computing application domain: naval ballistics firing tables
 Need: large memory, heavy-duty floating point
 Examples: CRAY T3E, IBM BlueGene
• Commercial: database/web serving, e-commerce
 Need: data movement, high memory + I/O bandwidth
 Examples: Sun Enterprise Server, AMD Opteron, Intel Xeon, IBM Power 6
Compsci 220 / ECE 252 (Lebeck): Introduction
28
More Recent Applications/Domains
• Desktop: home office, multimedia, games
 Need: integer, memory bandwidth, integrated graphics/network?
 Examples: Intel Core 2, AMD Athlon
• Mobile: laptops
 Need: low power, integer performance, integrated wireless?
 Examples: Intel PentiumM
• Embedded: PDAs, cell phones, automobiles, door knobs
 Need: low power, low cost, integrated DSP?
 Examples: Intel StrongARM, X-Scale, Transmeta TM3200
• Sensors: disposable “smart dust”
 Need: extremely low power, extremely low cost
Compsci 220 / ECE 252 (Lebeck): Introduction
29
Application Specific Designs
• This course is mostly about general-purpose CPUs
 CPU that can do anything, specifically run a full OS
 E.g., Intel Core i3, i5, IBM Power6, AMD Opteron, Intel Itanium
• Large, profitable segment of application-specific CPUs
 Implement some critical domain-specific functionality in hardware
o Graphics engines, physics engines
+ Much more effective (performance, power, cost) than software
+ General rule: hardware is better than software
 Examples: may or may not get to
o Emotion engine (PS2/PSX), Samsung phone, HDTV, iPod…
Compsci 220 / ECE 252 (Lebeck): Introduction
30
Why Study Computer Architecture?
• Understand where computers are going
 Future capabilities drive the computing world
 Forced to think 5+ years in the future
• Exposure to high-level design




Less about “design” than “what to design”
Engineering, science, art
Architects paint with broad strokes
The best architects understand all the levels
o Devices, circuits, architecture, compilers, applications
• Understand hardware for software tuning
• Real-world impact
 no computer architecture  no computers!
• Get a job (design or research)
Compsci 220 / ECE 252 (Lebeck): Introduction
31
Some Course Goals
• Exposure to the “big ideas” in computer architecture
• Exposure to examples of good (and some bad) engineering
• Understanding computer performance and metrics
 Empirical evaluation
 Understanding quantitative data and experiments
• “Research” exposure
 Read research literature (i.e., papers)
 Course project
 Cutting edge proposals
• Non-goals: “how computers work”, detailed design
• Let’s look at course home page…
Compsci 220 / ECE 252 (Lebeck): Introduction
32